I Stock 1251037767

To stave off the negative impacts of the COVID-19 pandemic, many clinical researchers pivoted away from traditional research pathways and harnessed novel research methodologies. One of these methodologies—machine learning—enabled researchers to conduct more socially-distanced data collection. Machine learning has aided continued positive developments of Alzheimer’s disease throughout the pandemic. Below, read about how these individual developments can work together to improve the patient experience.

Key Highlights

Learn about the developments in machine learning and Alzheimer’s disease research and how both can positively impact the patient experience.

Tweet
PRA Health Sciences
PRA Health Sciences

From the moment that the SARS-CoV-2 virus claimed its first victims, health scientists began a frantic investigation into the origins of the disease, its clinical features, its immediate consequences, and its long-term effects. However, the myriad of COVID-19 effects continues to evade the medical community.

COVID-19—with the majority of patients experiencing little to no symptoms and the minority becoming precipitously ill—has strained healthcare resources. In particular, clinicians caring for COVID-19 patients have struggled to correctly identify which patients presenting to emergency departments will experience severe acute kidney injury and require dialysis. The time-consuming need for specific equipment and hospital personnel for dialysis therapy limits the number of patients within a hospital who can simultaneously receive it as an intervention.

The Role of Machine Learning in Predictive COVID-19 Care

Innovative researchers at the Mount Sinai Health System turned to the power of machine learning. They recognized the challenge of predicting who within their care may need dialysis. Over nine months, they applied five machine learning models to analyze data from electronic health records (EHRs), predicting which patients may require dialysis based on data gathered from their first 12 hours of admission. They also used the machine learning models to effectively pinpoint the reason for a patient’s rapid clinical deterioration or death.

Building on their positive study results, the Mount Sinai researchers predict that the real-world application of machine learning can help hospitals more efficiently allocate resources and expertly manage patients by anticipating their future needs.

Machine learning has the potential to provide guidance ahead of time about a patient’s possible clinical course. This could make transitions more seamless and improve a patient’s overall hospital experience. This study is just one example of the future applications of machine learning. The possibilities are vast, particularly from a patient experience standpoint.

The Current State of Alzheimer’s Research

Alzheimer’s disease is a devastating diagnosis. It afflicts not only its victims, but their families as well. As the disease progresses, Alzheimer’s patients become less aware of their physical and cognitive deterioration—yet their care needs and the burden imposed on their loved ones both increase.

Unfortunately, Alzheimer’s disease is on the rise worldwide. The most recent report from Alzheimer’s Disease International estimates more than 50 million people live with dementia. This number is predicted to almost double every 20 years.

Researchers are currently investigating therapeutic modalities to help patients with Alzheimer’s disease across hundreds of clinical trials, 270 of which are supported by the National Institute on Aging (NIA). These trials vary in their stages of clinical development and their assortment of interventions, but all share a common goal of lessening the severe impact of this devastating neuropsychiatric condition.

However, the research world is experiencing renewed enthusiasm for Alzheimer’s care. The US Food and Drug Administration (FDA) recently decided to approve Biogen’s experimental drug, aducanumab, following two successful trials. This decision occurred within the context of other study results that show the Alzheimer’s funding shifting from drug companies to governmental and nonprofit organizations, further accelerating the development of innovative therapies. Among the many clinical trials underway, a few notable teams are investigating gene therapy and Alzheimer’s disease, the potential use of oral medications, and the impact of lifestyle measures.

Alzheimer’s Disease and the Role of Machine Learning

Machine learning is advancing Alzheimer’s research, similarly to the ways it accelerated predictive care for COVID-19. Several clinical trials are currently underway that capitalize on the power of artificial intelligence and machine learning within the field of dementia research.

Here are three current examples of how machine learning is advancing Alzheimer’s care:

AI-based Fall Detection Technology

Researchers at a US-based trial out of the UC Berkeley AI Research Lab, SafelyYou, are examining the impact of an AI-based fall detection technology for residential communities that specialize in long-term dementia care. Falls represent the most common incident that necessitates hospitalization for Alzheimer’s patients. The researchers note that, due to COVID-19 distancing measures, falls in memory care have increased by 20%. Their technology detects falls using deep learning artificial intelligence algorithms paired with wall-mounted cameras. A human-in-the-loop (HIL) at a central call center confirms a fall and alerts residential staff, and an occupational therapist then reviews the fall video to make recommendations for how to arrange the physical space to prevent future falls. In this way, machine learning helps prevent falls from happening in the first place. It also provides immediate assistance to patients who do experience a fall event, which significantly improves their care experience.

Early Alzheimer’s Detection

Researchers at a trial based out of Indiana University estimate that while the societal cost of Alzheimer’s disease is high, half of Americans living with Alzheimer’s Disease and Related Dementias (ADRD) never actually receive a diagnosis. Early Alzheimer’s detection is helpful for patients and their families to make preparations and anticipate future needs and may help patients stop the progression of their dementia. Similar to the machine learning applied to the EHR used by researchers for predictive COVID-19 care, these researchers integrate EHR information from patient annual wellness visits with machine learning models to develop a cheap and scalable method of early ADRD detection.

Life Story Work

Researchers based in Ohio note that many non-pharmaceutical interventions can positively impact the course of Alzheimer’s disease, particularly an intervention known as life story work. Life story work is defined as “the use of written and oral life histories.” However, as the Alzheimer’s patient population grows, conducting effective life story work is more difficult for staff members of residential long-term care facilities. To address this challenge, these researchers evaluate a machine-learning-based platform that generates life story materials by converting speech to text to help patients engage in reminiscence therapy and share their stories with their caregivers.

As the field of machine learning advances, Alzheimer’s patients only stand to gain from these innovations. Clinical research teams at PRA are committed to furthering research on this devastating condition.

The Impact of Machine Learning on the Patient Experience

Machine learning represents a helpful ally for clinicians and researchers alike as they attempt to aggregate large amounts of clinical data from disparate sources to find meaningful patterns. However, patients themselves stand to gain the most from machine learning implementation within the real-time clinical and research worlds.

For example, in addition to predictive COVID-19 and Alzheimer’s care, machine learning has also been shown to help patients navigate their risk of devastating gastrointestinal bleeds after starting blood-thinning medications, outperforming other risk modeling methods. Additionally, researchers have shown that machine-learning models may help clinicians anticipate blood clots in pediatric patients, a risk assessment that could dramatically impact morbidity, mortality, and the overall patient experience.

Takeaways for Researchers and CRO

Machine learning is already enhancing the healthcare world, and its positive impacts extend to the clinical research realm. Sponsors and CROs searching for better, more impactful ways to analyze existing data, isolate new patterns, and even collect better data should consider partnering with machine-learning models. Machine learning can enable research teams to simplify their data acquisition and data synthesis. For study participants, this can mean that machine learning effectively enhances research findings and amplifies the value of their contributions and, potentially, personal outcomes.


The more research participants get cared for in a deliberate, data-driven manner, the more likely they are to feel empowered, engaged, and satisfied with their trial participation and clinical outcomes. The clinical teams at PRA are committed to driving innovation forward globally, and machine learning can be a powerful asset in this mission.

Learn what our experts have to say on machine learning

Read more

Explore related content and topics